Google Tests Gemini Skills Kill Switch in Chrome Canary After Silent 4 GB Model Download Sparked User Backlash
Google is testing a dedicated on/off toggle for Gemini Skills inside Chrome Canary, giving users a separate control from the browser’s broader AI settings for the first time.
The switch appears under a new “Skills in Chrome” section within Chrome’s “AI in Chrome” settings page, sitting alongside existing options such as “Help me write” and the main “Gemini in Chrome” assistant controls.
What Triggered the Move
The toggle follows a wave of user complaints after Chrome quietly downloaded a roughly 4 GB copy of Gemini Nano — a locally run AI model — onto Windows and macOS machines without explicit user consent.
Many users discovered the download only after their available disk space dropped unexpectedly or system resource monitors flagged unusual activity.
The backlash pushed Google to address growing concern that Chrome’s AI features were expanding too aggressively and too automatically in the background.
What the Toggle Actually Does
Switching “Enable Skills in Chrome” off immediately disables Gemini Skills — the feature That Lets users save and rerun prompts for repetitive in-browser tasks.
The Skills gallery stops loading and instead displays a message that “Skills aren’t available” because the feature has been disabled in settings.
That said, turning off Skills does not Shut Down everything. The Gemini side panel remains accessible, and other Gemini in Chrome features continue to function normally.
The new control is also separate from Chrome’s existing on-device AI toggle, which governs the large local model download itself.
By splitting Skills into its own switch, Google gives users finer-grained control over which parts of Gemini actually run — without forcing them to disable the entire AI layer.
Still an Experiment
The toggle surfaced in testing on Chrome Canary v150.0.7871.2, the browser’s most experimental release channel. It does not yet appear in Chrome’s stable or beta channels.
Even within Canary, the setting does not show up for every user — only those running the latest build with AI features enabled can see it.
That limits its immediate reach. Canary attracts a small fraction of Chrome’s global user base, which StatCounter estimates holds roughly 65% of the worldwide desktop browser market.
Other Experiments Running Alongside It
Google’s Chrome team is also testing separate changes unrelated to AI controls.
Engineers are evaluating which glass visual effect to apply to an upcoming Windows UI refresh, and they are experimenting with individual alignment settings for the browser’s side panel.
A feature internally called “Indigo” is also in testing — it would allow users to replace images on webpages with AI-generated alternatives.
None of those features have reached stable Chrome, and Google has not announced a timeline for any of them.
Gemini Nano, the on-device model at the center of the storage complaints, is a compact version of Google’s Gemini AI family designed to run locally on consumer hardware rather than relying entirely on cloud inference.
